Satellite lunch symposium 2019

AI, Deep Learning and Multi-Parametric Assesment for Advanced Imaging and Diagnostics

Date: Wednesday, February 27
Time:
12:30 – 13:30
Room:
 K (Lower level)

Chair person: Prof. Catherine Oppenheim, MD, Centre Hospitalier Sainte-Anne, Paris, France

Click on the image to see the details.

Speakers

Roadmap for Artificial Intelligence in Radiology: Recognition, Reconstruction, Reasoning

Prof. Bram van Ginneken, PhD, Radboud University, Nijmegen, The Netherlands

Artificial Intelligence, in the form of deep learning, is extremely successful in automated recognition of abnormalities, in segmentation and in quantification. I will illustrate this with some recent results. Increasingly, deep neural networks are used to generate new images. They can be used for reconstruction, for denoising extremely low dose scan, for creating novel images, and for mapping images from one modality to another, for example CT to MR. The next frontier is to go beyond mappings from input data to output data and design systems that can reason based on variable inputs.


Deep Learning Reconstruction: The Next Step in CT Image Quality

Prof. Matthias Prokop, MD, Radboud University, Nijmegen, The Netherlands

The first clinical results of Deep Learning Reconstruction (DLR) algorithm for CT (both 320-row CT and UHR-CT) will be presented, featuring a deep learning neural network that can differentiate and remove noise from signal, creating extraordinary high quality images.


MRI and Advanced Algorithms like Computed MRI from Research to Clinical Practice

Prof. Luca Saba, MD, University of Cagliari, Cagliari, Italy

Computed MRI finally has become more practical in a clinical setting. Clinical imaging therefore may very well profit from these advanced imaging techniques. Where in the past these tools were mostly used in research it can also provide flexibility and reduction in examination time in daily practice. It also offers the possibility to offline reconstruct conventional sequences with different contrast settings and it allows to obtain quantitative values that may be used for follow up or for tissue characterization.


Multi-Parametric Approach for Diffuse Liver Disease with Ultrasound

Prof. Valérie Vilgrain, MD, Hôpital  Beaujon, Clichy, France

Artificial Intelligence, in the form of deep learning, is extremely successful in automated recognition of abnormalities, in segmentation and in quantification. I will illustrate this with some recent results. Increasingly, deep neural networks are used to generate new images. They can be used for reconstruction, for denoising extremely low dose scan, for creating novel images, and for mapping images from one modality to another, for example CT to MR. The next frontier is to go beyond mappings from input data to output data and design systems that can reason based on variable inputs.